A federal model can now be called "super intelligence" in a new agency memo and "artificial intelligence" in the statute that memo cites. That split is already official. Executive Order 14434 tells executive agencies to use "Super Intelligence" and "SI" in new correspondence, websites, reports and policy documents, while defining those terms through the existing legal definition of AI. The White House has 60 days from September 29 to propose a replacement definition. A newly announced Super Intelligence Force has twice as long to report on the technology's risks and opportunities.
President Donald Trump announced the force on October 4. Director of National Intelligence Jay Clayton will lead it while keeping his intelligence post, according to TechCrunch's account of the announcement. FTC Chair Andrew Ferguson, Undersecretary of War for Research and Engineering Emil Michael and Office of Personnel Management Director Scott Kupor will serve as vice chairs. Politico reports that the group will report directly to Trump and White House Chief of Staff Susie Wiles.
The task force's reported charter gives it 120 days to assess risks and opportunities, develop plans for threats enabled by the technology and consider how to avoid both excessive regulation and regulatory capture. Those assignments are broad. The narrower definition deadline comes first, and it will decide what the government means when it puts the SI label on a procurement or security rule.
A new label still points to the old law
The September 29 order changes executive-branch language immediately, to the extent the law permits. It does not create a new technical threshold for SI. Section 3 says SI covers the technologies already included in the AI definition at 15 U.S.C. 9401(3). That law describes a machine-based system that pursues human-defined objectives by making predictions, recommendations or decisions that affect real or virtual environments.
That scope reaches far beyond a hypothetical system smarter than a human at every task. Under the order's current wording, an agency can use SI for a qualifying recommendation engine, predictive system or automated decision tool. A vendor should not read the new name as a frontier-model exemption for smaller systems. Until another presidential action or an act of Congress changes the definition, the order expressly keeps the statutory AI scope in place.
The Assistant to the President for Science and Technology must submit proposed legislative language within 60 days. The proposal has to assess whether SI should modify, expand or replace the existing definition, identify amendments needed for statutory references to AI, and recommend any further executive action. This is the first substantive deadline. It asks the administration to decide whether the rename remains an alias or becomes a different legal category.
Executive Order 14434 also draws a boundary around the find-and-replace job. Agencies do not have to alter previously issued regulations, contracts, grants or presidential actions. New non-statutory documents should use SI, but an active contract can retain AI. The same project may therefore carry one term in its signed paperwork and another in later agency communications. That is an administrative mismatch, not evidence that the underlying model changed.
The mixed vocabulary is already visible
NIST offers a useful snapshot of the transition. Its relevant team is now the SI Standards and Guidelines Group, and the page says the group develops standards for SI data, performance and governance. On that same page, its central publication remains the "NIST AI Risk Management Framework (AI RMF)." The linked framework page still describes AI products, services and systems, and says version 1.0 is under revision.
Federal buying guidance has the same legacy vocabulary. The Office of Management and Budget's April 2025 memorandum is titled "Driving Efficient Acquisition of Artificial Intelligence in Government". It tells covered agencies to review proposed AI purchases, examine performance and risk practices, and obtain documentation that supports post-award monitoring. Because the new order exempts historical documents, those instructions do not lose effect merely because their title says AI.
For developers and vendors, the safe reading is literal: map SI references back to the governing definition and cited policy instead of guessing from the new phrase. A fresh solicitation may ask for an SI system while incorporating the statute alongside OMB or NIST material that says AI. Scope comes from those definitions and contract terms. The branding alone does not answer whether a common commercial feature is covered, whether a use is high impact or what testing an agency expects.
This also matters outside the United States. The NIST standards group participates in international standards work, where software providers reuse model cards, risk registers and compliance evidence across customers. US federal material will now introduce SI into documents that may also cite standards written around AI. Maintaining an explicit term map is mundane work, but it is less risky than silently treating SI as a narrower or more advanced class of system.
Clayton inherits work already on a 120-day clock
Putting the director of national intelligence in charge makes security central to the force's structure, but the administration had already assigned Clayton's office a long list of AI jobs. A June national security memorandum gave the DNI, defense officials and the NSA 120 days to produce baseline security practices for national-security AI systems and standardized methods for test, evaluation, verification and validation. It also ordered reviews of procurement, model exchanges and technical training on the same timetable.
The June memo uses the AI label and the same statutory definition now inherited by SI. It covers the national security enterprise rather than the whole federal government, and it assigns named deliverables to existing officials. The new force's reported charter is wider and less specific. Its first report will need to explain how it relates to those projects and whether any separate track remains. Otherwise, two 120-day processes led in part by the DNI could issue overlapping advice under different names.
The other vice chairs widen the set of agencies at the table. Ferguson brings the FTC's competition and consumer-protection authority. Michael oversees defense research and engineering. Kupor runs the federal personnel agency, which the June memorandum already tasked with developing a reserve of outside AI talent and a national-security curriculum. Their presence identifies the offices involved. The announcement does not by itself give the force new rulemaking power or replace those agencies' existing authorities.
That limit matters when reading the charter's reported goal of addressing threats while avoiding overregulation and regulatory capture. A report can recommend policy or legislative action. Binding rules still need a lawful route through Congress, a regulator or presidential authority. Executive Order 14434 preserves agency powers, OMB functions and applicable law, and it makes implementation subject to available funding.
The report needs measurable boundaries
The task force can make its 120 days useful by publishing the definitions and handoffs that the announcement leaves open. Start with the scope of SI. If the administration keeps the broad 2021 AI definition, its report should say so plainly and avoid language that implies every covered system has beyond-human abilities. If it proposes a narrower class, agencies and contractors will need criteria they can test rather than a promotional label.
Ownership needs the same precision. The force should identify which existing work products satisfy parts of its charter, including the national-security testing methods and risk guidance ordered in June. It should also state which agency handles a reported model failure or a procurement and competition dispute. Naming senior officials creates a coordinating table. It does not tell a developer where evidence should go.
The current federal acquisition memo already asks agencies to test products in conditions close to their intended operating environment and to track performance after an award. The task-force report should connect any proposed control to a failure mode and a measurement method, then name the systems in scope. A new acronym cannot substitute for those details.
The first document to watch is the proposed federal definition due in late November, followed by the force's report roughly two months later if its reported 120-day clock began at launch. Also watch whether the White House publishes the charter itself and how NIST and OMB revise their mixed AI and SI vocabulary. A developer should be able to trace an SI requirement to a defined system and a measurable check owned by a named office. Until that mapping exists, the government's new name is moving faster than its policy machinery.